|
Predictive Analytics: from CRM to EDM
BY KENNETH CLINE AND PAUL McADAM
Fair Isaac?s Mark Greene says the key to customer profitability lies in transaction data.
|SYNOPSIS | Fair Isaac Corp. is best known in the financial services industry for its ubiquitous FICO credit scores, which are used by credit bureaus and banks to judge the credit worthiness of consumers. But CEO Mark Greene sees more of the company?s future in Enterprise Decision Management (EDM) or predictive analytics that mine customer transaction data for information that can help banks improve their marketing capabilities. Despite the financial services industry?s past disappointments with Customer Relationship Management (CRM), which also sought to use customer information to improve profitability, Greene says EDM can show a true bottom-line payoff—and advance the frontiers of payments.
Mining customer transaction data for marketing purposes may strike many bankers as a kind of ?90s thing.? Disappointments with trying to implement Customer Relationship Management (CRM) strategies during that decade still loom large in the industry—been there, done that, thank you very much. But Mark N. Greene, CEO of Minneapolis-based Fair Isaac Corp., wants financial institutions to take another look at the data mining concept under the label ?Enterprise Decision Management? or EDM. Where CRM was backward looking, telling bankers what had happened in the past, he asserts that EDM can look forward, incorporating predictive analytics that can anticipate what customers will want to buy.
Time will tell whether EDM becomes a buzzword in this decade similar to what CRM was in the 1990s. But Greene, who will be speaking at the BAI TransPay Conference & Expo in February, strongly encourages bankers to consider EDM and predictive analytics as a potential new frontier in payments. To a certain extent, Fair Isaac?s own future is riding on the result. The company?s earnings have been essentially flat since 2005 and one prominent shareholder has been agitating for a sale. Greene himself was brought on board in February 2007 to chart a new direction for a company whose main product—the FICO score—was first developed fifty years ago. Since then, Greene has grappled with internal issues such as a sales reorganization and a lawsuit against the credit bureaus involving VantageScore, their alternative product to FICO.
To gain some forward momentum, as he explains in the following interview, he?s looking to EDM, along with expan-sion overseas and in industries outside of financial services.
Q: How do we get from Customer Relationship Management (CRM) to Enterprise Decision Management (EDM), considering that banks had some negative experiences with CRM? How is EDM different?
Greene: The relevant part of CRM is the part that’s better known as “business intelligence” or the state-of-the-art practice for banks to mine transaction data and to understand their customers. A good CRM system has imbedded within it a business intelligence system that lets you ask questions such as: Who are my most profitable customers? What’s my profit rating in Minneapolis? Those are all valuable capabilities, but they’re like looking in the rearview mirror—they only tell you what happened.
EDM evolves from a similar concept. But instead of looking back, it looks forward and is predictive. It asks: To whom should I try to sell what products? Is this applicant standing at the door worthy of this mortgage? Who will be my most profitable customers over their life cycle if I do X, Y and Z? So, the analytics of EDM are very similar, but they?re applied in a modeling and predictive sense, as opposed to a rearview mirror sense.
Q: Still, given industry skepticism, how can you convince banks that EDM will bring the benefits promised?
Greene: First, where CRM fell short was not so much in the models and the data accumulated, but in the failure of organizations to imbed it in their processes and train people how to use it. It’s one thing to build great computer systems; it’s quite another to ensure they are used by motivated training teams. That will be true as well for EDM.
Second, there’s a clear business benefit with EDM. The kind of results we’re seeing from these EDM models and the potential for business improvement are dramatic. We’re seeing organizations able to add extra products to their share of wallet, moving for example from four products to six. We’re seeing the lifetime profitability of relationships go up 30% and 40%. We’re seeing lifetime reductions in fraud in a similar range, 30-40%.
The problem with CRM is that it looked good on a whiteboard, but people didn’t really see the “there” in the numbers. With EDM, we’re starting to see the “there” in the numbers.
Q: Which categories of banks have you found most receptive to EDM?
Greene: It’s usually the larger banks, the more sophisticated ones that tend to have a good data mart already in place. You can’t do EDM if you don’t have a good data repository.
You also have to have an appreciation for the power of the math and analytics. That’s particularly true of banks that have a lot of different product lines and very broad customer sets, where they have a cross-sell challenge.
We offer our clients a matrix that places their customers in one dimension and their products in another. We model which customers have which products. It’s always a distressingly sparse matrix. There aren’t too many people buying a broad range of offerings from their bank. EDM can help banks connect the dots and connect more customers with more products.
So, EDM resonates with tier one and, to a certain extent, tier two banks in the U.S. and then all the global banks. Beyond that, we’re seeing interest in a couple of other related industry verticals, including retail and telecommunications.
Q: In your 3Q07 conference call, you listed several European banks as being receptive to EDM but no U.S. banks. Why?
Greene: It has nothing to do with the U.S. banks’ adoption of EDM but everything to do with their willingness to be publicly cited as a commercial partner. I think I can safely say that all of the top five U.S. banks are well down the EDM road with us. But they view it as a source of competitive differentiation so they’re not eager to tip off their competitors to what they’re doing. Pure and simple, I wish it were otherwise. It would be great for our business if we could talk about it more. But we respect our clients’ desire to protect a competitive edge.
Q: There’s a lot of talk in the industry about achieving a single, or 360 degree, view of the customer, as opposed to the fragmented view created by bank silos. Does EDM do that?
Greene: Yes it does. What we don’t require in our approach is that all the data live in one place. And it typically does not. So, we’re pretty good at deploying EDM in environments where we have to pull and mix-and-match from multiple sources and collate the data. But when we’ve completed that, we do end up with a 360-degree view of the customer. And the closer we can come to that, the better our EDM models do in helping banks to improve cross-sell possibilities and reduce risks.
Q: Looking behind the concept, what technological advances make EDM feasible?
Greene: In the last five years or so, the industry has become very sophisticated at using analytics that we term “data spiders.” Basically, they’re transactional analytics, analytics that look for patterns across transactions, which allow us to infer your preferences, your needs and your next actions.
Let me give you an example of how this works for a retailer. Suppose you were in the store last week and bought an i-Pod and then came back this morning and bought a microwave oven. Our model looks at that kind of transactional data, data from other consumers similarly situated and at your past transaction history. All of this data might suggest a 40% likelihood that you’d buy a flat screen TV next Thursday if you were given a coupon for $100 off.
That 40% accuracy level compares with conventional results of closer to 4%. Retailers today are thrilled if they mail out a coupon and get a 4% hit rate. That’s nirvana. The kinds of models that we’re talking about here can hit 40%, a 10-fold improvement in predictive accuracy.
The problem gets a little bit harder to solve when consumers operate in a non-closed shopping environment, where you visit many stores and use different credit cards for your purchases. Somebody has to sit in the middle of all those transactions and figure it out. We don’t do that because we don’t own any of that data. But we provide the tools for organizations that can do exactly that. So, modeling customer behavior across multiple channels and multiple payment instruments is an increasingly fertile field.
Q: Is this modeling activity mostly confined to the card domain at this point?
Greene: No, although that’s the most common place. The prerequisite for this to work is to possess relatively clean—rather than noisy—data. Both credit and debit card data tends to be very clean. By contrast, cash purchase records are pretty noisy for many reasons. Retailers don’t like to share exactly how much they have in the till.
So, it’s predominately in the card domain today, but we’re looking at other forms of payment.
Q: What are the difficulties or obstacles within banks to making use of this data?
Greene: They are two-fold. The first is fairly straightforward to solve, if given proper attention, and that’s getting the proper consumer privacy issues addressed. These tend to be opt-in situations, where customers must have adequate notice and provide informed consent. It’s driven by regulatory initiatives.
The second challenge is silo complexity within financial institutions. We’re talking about accumulating transactions across many different product lines, across many different silos. The “E,” or “Enterprise,” idea in “EDM” is really important, but it’s also still difficult for many traditional banks to solve.
A couple of banks are well down that road. I’m not giving you any customer secrets here, but Charlotte-based Wachovia Corp. and Capital One Financial Corp. of McLean, Va. are widely known as well-integrated organizations. Those kinds of institutions have an ability to execute EDM pretty well.
We’ve been talking here about using transaction data in a positive sense, for predictive modeling. It’s also possible to use EDM in a negative sense, to avoid problems. I’ll give you an example that we can talk about publicly.
Royal Bank of Scotland uses some of Fair Isaac’s software in their credit card collections and recovery area. The software suggests which transactions the bank should focus on trying to collect and which receivables are really not worth the effort because they’re fraudulent transactions that can never be collected. The bank asked us if it would be possible to take that knowledge of fraud in the back office, on the back end, and plug it into the beginning of the credit life cycle, in what’s called the “originations” phase, so they could avoid opening bad accounts. You can reduce fraud by 50% if you take that kind of approach.
So that’s one use of EDM for which Fair Isaac is well known in the industry.
Q: Let’s talk about the company itself for a moment. Fair Isaac’s earnings have been sluggish for the past couple of years. Most recently, you’ve struggled with a sales reorganization. When do you expect to see improvement?
Greene: We’re actually starting to see improvement right now, as reflected in our fiscal 2007 results.
The markets in which EDM applies are growing at least 7% to 10% a year. As one of the leaders in the field, you’d expect that we should be able to grow at those rates too. Instead, we’ve been essentially flat for eight or nine quarters. But that does not speak to the inherent growth prospects of the company in my view. The inherent growth prospects are much closer to that market growth rate.
We’ve now addressed operational issues around sales discipline, service discipline and ensuring client satisfaction. Our flat growth trajectory was not a defect of the products, nor was it a defect of our people—it was very much a defect of how we did business. I’m actually quite encouraged that we’ve made a lot of progress. We’re beginning to see it show up in repeat business and larger-scale commitments from customers.
Q: Speaking of defects in how the company did business, you’ve stated publicly that Fair Isaac had a reputation in the industry as somewhat arrogant. What are you doing to repair that bad rep?
Greene: Having worked in research organizations through the early part of my career, I know that arrogance can sometimes be viewed as a good signal, in the sense that it means we have really smart people who are very clever at figuring out complicated problems. When you look at the complexity of the stuff our folks work on and the powerful results they are able to achieve, you can’t help but be impressed.
Unfortunately, being impressed by our own capabilities led us to dictate how people should do business with us, as opposed to going out and listening to our customers. This is a setting-the-tone-at-the-top issue and it’s about reinforcing cultural values. I’m spending a lot of time reinforcing the importance of our top core value, which is focusing on clients. We’ve done third-party client satisfaction surveys and established a council for our top thirty customers. We also maintain a dynamic list of our top ten “most unhappy” customers and we quickly figure out how to make them happy customers. We’re now starting to hear customers talk about the “formerly arrogant Fair Isaac.”
I’ll give you one example of the progress we’ve made. Soon after I arrived here, I visited a large card organization where the chief information officer (CIO) greeted me with, “I’ll give you three minutes before I kick you out of my office.” We had treated that organization rather abusively.
That three-minute visit turned out to last an hour and a half. When I called a few weeks ago, just to see how things were going, the CIO began the conversation by saying, “I’ll give you three minutes before I hang up on you.” I said, “Oh, no, what now?” His answer came back: “Because three minutes is all it’s going to take to describe what a wonderful job Fair Isaac is doing now in taking care of us.”
I’m hearing more and more of that, which means we’re heading in the right direction.
Q: There’s a lot of consolidation right now in the third-party services industry. What do you think is driving that and what’s Fair Isaac’s stance?
Greene: There are three classes of players converging on the analytics and decision technology space. There are transaction processors, like Fiserv. There are credit processors, like the credit bureaus. And then there are predominately software houses, which would include us, SAS and SAP. At least two of those vectors, credit processing and transactions, are scale businesses; you win by having economies of scale and by getting larger. The software business has some of that. But the place where we play in software, which is characterized by high analytic value content, is less prone to that kind of consolidation. It’s less commoditized but it’s not immune to consolidation either.
So our strategy is to do two things. One, we heavily emphasize innovation. We will win in this business by being a step or two ahead of the other guys. Our FICO score, for example, which is the thing we’re best known for, is refreshed every eighteen months.
Second, we grow to achieve a certain scale. A good portion of what I’ve been doing since arriving here is pushing the company into both new markets, such as China, and into industry verticals, such as retailing. If you get to a certain critical mass, you’re less of a target for acquisition. And our board is quite supportive of that.
Q: How does the recent turmoil in the credit markets affect your business? Does it help or hinder?
Greene: If the turmoil were to persist long enough, there would be a chilling effect on all kinds of spending. People would just pull in their horns—and their Information Technology budgets. That has not yet happened because the turmoil has not lasted long enough. There are certain segments, like subprime, that have continued their aggregate level of spending because they’re using more analytics to make smarter decisions in the face of uncertainty.
Q: So you benefit on the risk or fraud side?
Greene: Fraud and the use of FICO scores. In subprime, our business has done well in recent months because of the uncertainty and danger in those markets. We’re a company that helps people stay out of trouble. So when times are risky, so long as they’re not calamitously risky, we tend to do fairly well.
Q: How did your experience at IBM prepare you for this particular position?
Greene: First of all, I’m actually a banker by background. Before joining IBM, I worked at Citibank and the Federal Reserve. I ran modeling and forecasting groups at those two organizations so I understand the business that bankers are in. I certainly understand our customers.
IBM did teach me a few things. For example, I learned the importance of focusing on customers because they pay the bills. IBM also taught me the importance of prioritization and clarity in a matrix organization. At Fair Isaac, like IBM, we’ve got different industries, different countries and different products, albeit on a much smaller scale.
I guess the last thing that’s relevant, especially to our current strategy, is that IBM taught me the importance of having a global view. About 70% of our business is North American: U.S. and Canada. The other 30%, the international part, never got quite the attention it deserved but does now. We’re recognizing distinct opportunities in China, Japan, Eastern Europe and Brazil.
Q: What sort of things has Fair Isaac learned from its international work that can be applied to U.S. banks?
Greene: European banks tend to have a pretty sophisticated view of risk management in a Basel II sense. That’s a lesson that some U.S. banks have figured out, but not a lot. So, best practices and Basel II compliance tend to come from outside North America.
European banks also are leaders in terms of attaining that 360-degree view of the customer. Not because of better technology or processing, but maybe because of superior cultural understanding, they do a better job of wrapping themselves around their customers. We look for know-your-customer best practices to come out of Europe.
Q: And a growing portion of your revenue stream is coming from outside banking, retail and healthcare, for example. Can any of those best practices be brought back into banking?
Greene: I’d put it the other way around. Financial services companies tend to be more advanced in their use of analytics—data mining and business intelligence. So we typically show things to retailers and health care providers derived from financial services.
There is one big exception. Part of our retail work is in the hospitality industry, which is very good at customer care. They definitely have a 360-degree view of the customer.
|
